Pay me Right: Reference points and Executive

Compensation

Aleksandra Gregoriµc� Sa�o Polanecy Sergeja Slapniµcarz

April 8, 2009

Abstract

This paper studies the impact of external reference points on the CEO compen-

sation. We examine how a change in the reference value in�uences the compensation

contracts of new managers and how these values a¤ect the adjustment of the exist-

ing pay contracts; for the latter we con�rm that compensation is rigid downwards.

In line with the managerial power theory of executive pay, we also �nd that upward

adjustments are on average stronger in the �rms with more dispersed ownership

and thus, stronger CEO power.

Keywords: Executive compensation, bargaining, reference points, ownership

structure.

First draft. Please, do not quote.

JEL Classi�cation Numbers: G30, G34

1 Introduction

Executive pay practices have been in the centre of the academic interest for more than

two decades. Among the many practices, the one that still generates lots of ambiguities is

the use of compensation consultants, peer group comparison and external salary surveys

to calibrate managerial pay (Wade, Porac and Polloc, 1997). On the one hand, such

benchmarking may improve compensation contracts in the way that it helps the Board

�Copenhagen Business School, Department of International Economics and Management, Centre forCorporate Governance and, ECGI. E-mail: agr.int@cbs.dk.

yUniversity of Ljubljana and Institute for Economic Research. E-mail: saso.polanec@ef.uni-lj.sizFaculty of Economics, University of Ljubljana. E-mail: sergeja.slapnicar@ef.uni-lj.si.

1

of directors determine the labor marker conditions and gauge the value of the CEO�s

reservation pay (Holmstrom and Kaplan, 2003). However, a number of studies challenge

this view by showing how the CEOs can in�uence the selection of the peer group (Wade,

O�Reilly and Chandratat, 1990, Murphy, 1999; Porac, Wade and Polloc, 1999); these

studies suggest that peer-group comparison and other benchmarks mostly serve to re-

anchor the bargaining process that is, to introduce a new reference value in the pay

bargaining between the Board and the CEO. This paper contributes to the existing

literature by systematically exploring the impact of CEOs�reference values on the level

and structure of their pay.

From the perspective of the CEOs, the reference values help them determine whether

they are fairly paid (Falk, Fehr and Zehnder, 2006). That is, the CEO evaluates the

appropriateness of her compensation by comparing her reward to a reference value; the

latter is determined on the basis of the rewards of comparable persons within the orga-

nization or industry or, upon the CEO�s expectations and aspirations (Wade, O�reilly

and Pollock, 2006; Kahneman, 2002). Despite the recognition of the importance of the

references and benchmarks in the executive compensation setting, the role of reference

points is not yet fully understood. Can reference points drift executives�pay upwards?

Can a change in one party�s reference point motivate the re-negotiation of a contract

and in�uence the subsequent bargaining outcomes? According to the classical bargaining

theory, only a rise in one party�s outside option or a change of its bargaining power can

trigger the re-negotiation of the pay. Recently however, the importance of the reference

values for bargaining, contracts and behavior has been gaining recognition in the liter-

ature. Compte and Jehiel (2003), for example, de�ne reference points as the outcomes

that the parties reach in the previous bargaining stages and theoretically model the way

these references impact the subsequent bargaining process. In Hart and Moore (2006),

the need to frame the agents�reference values and avoid shading and retaliation increases

the e¢ ciency of the rigid contracts. A reduction of the agents�utility and e¤ort due to a

divergence between their reference values and contract terms (i.e. rewards) has been also

addressed in Akerlof and Yellen (1990) and Koszegi and Rabin (2006). In the setting of

executive compensation, we would thus expect that any remuneration below the CEO�s

reference value will aggravate her utility, to which the CEO will respond by adjusting

her e¤ort downwards or by re-negotiating better terms with the Board. Apart from

2

preventing a decline in the CEO�s e¤ort, the literature advances several other reasons

why the Board of directors may be sensitive to the CEO�s reference levels and willing to

re-negotiate and adjust the executive compensation accordingly. Managers gain power

over the Board through persuasion and selective use of information; they can in�uence

the Board by selecting outside board members, granting them bene�ts and favors (i.e.

rent-sharing) or developing social relationships with them (Wade et al., 1990; Buchholtz,

Young and Powell, 1998).

To analyze the impact of reference points on executive compensation, we draw on

the evidence of a new European country, Slovenia. This country is interesting since, at

a point during transition, Slovenian managers outlined the guidelines or, the so-called

Criteria for executive compensation, through which they successfully advocated a rise

in their pay. By re�ecting what the managers at that time considered the appropriate

reward for their work, this document provides us with a valid and accurate measure of

the CEO�s reference value. This is an advantage of our study since the identi�cation of

the reference value in general represents a compelling research problem. Given the fact

that reference points are based on individuals�perceptions (Rizzo and Zeckhauser, 2003,

Ezzamel and Watson, 2002), the validity of research �ndings depends on how well the

researchers de�ne and measure the reference points. No such concern applies in our case.

We provide new and clear evidence on how the managers�reference value drift the levels

of executive compensation upwards. The impact is observed for both the newly stipulated

compensation contracts and, for the existing contracts. For the latter, the adjustments

towards the reference compensation are rigid downwards. In addition, we �nd that the

reference value exhibits a relatively stronger impact on the upward adjustments of the

CEO pay in the joint stock corporations (in comparison with limited liability �rms); in

these �rms, the ownership is more dispersed and consequently, the power of managers�

in the pay bargaining process stronger.

Our study adds to the previous literature in many important aspects. First and most

importantly, we provide unique evidence on how the CEOs�reference values in�uence the

stipulation of the pay contracts and consequently, the level and structure of the executive

pay. Second, the circumstances that facilitated the adoption of a new reference for the

CEO compensation and the di¤erences in the adjustment across alternative corporate

forms support the managerial power explanation of the CEO compensation (Weisbach,

3

2006; Bebchuk et al., 2002). Finally, we contribute to the scarce research on the execu-

tive pay in other parts than Western Europe and the USA. Contributions from transition

economies, in particular, are minor (i.e. Eriksson, 2005; Jones and Kato, 1996). Ex-

ploring how executives are remunerated in di¤erent social and organizational context

is a warranted contribution to the general understanding of the compensation practices

(Zajac and Westphall, 1995).

The remainder of the paper is organized as follows. The second section provides

a brief overview of the relevant literature. The third section describes the institutional

context in Slovenia and the design of the reference points. The fourth section presents the

model speci�cation, measurement and estimation issues and hypotheses. Data, sample

and descriptive statistics are presented in the �fth section. The sixth section presents the

empirical results. Last section concludes.

2 References points

In this section, we introduce the term "reference point" and present a short overview

of the relevant theoretical and empirical �ndings. Di¤erent de�nitions of the reference

points or reference values1 can be found in the literature. Koszegi and Rabin (2006)

for example, de�ne the reference point as a person�s probabilistic belief about a relevant

variable, which is in turn determined upon the expectations about an outcome that a

person had in a recent past. In the bargaining setting, Li (2004) and Compte and Jehiel

(2003) suggest that reference points evolve endogenously, as o¤ers in the prior bargaining

phases. Hart and Moore (2006) de�ne reference values by a range of possible outcomes

that the parties determine with a contract. In relation to personal income, Rizzo and

Zeckhauser (2003) consider reference points as the �desired�or �target�level of individual�s

earnings. These are in turn determined by the individuals�aspirations or, upon their

comparison with the salient others. However, the psychological literature o¤ers little

guidance as to which reference group can be considered as relevant in this comparison

(Akerlof and Yellen, 1990). For the purpose of empirical investigation, reference groups

have been constructed as groups consisting of all the individuals living in the same country

(Easterlin, 1995), as groups of people within the same profession, age and employment

status. Finally, reference points may arise as a response to an external event (c.f. Perry

1In this paper, the terms reference point and reference value are interchangeable.

4

and Zenner, 2001, Rose and Wolfram, 2000). An example of an exogenous reference

point is the tax exemption of �xed pay over one million US dollars in the USA. This

regulatory act signi�cantly anchored the pay bargaining at the �rm-level and (contrary to

the expectations) led to an increase in the executive pay; in fact, once the regulator de�ned

$1 million compensation as reasonable, many companies below the threshold increased

their cash compensation to $1 million regardless of underlying economic circumstances

(Rose and Wolfram, 2002).

A number of studies show that people�s perceptions are "reference dependent" (Kah-

neman, 2002). A departure from a reference point introduces a �gain-loss�perspective

of an individual�s utility (Koszegi and Rabin, 2006). In other words, the utility that an

individual associates to a given outcome (�perceived utility�) is determined by both, the

outcome and its relation to the reference point. References in�uence people�s aversion

towards risk. According to the prospect theory the utility function breaks at the reference

point and is considerably steeper for losses than for gains (Kahneman and Tversky, 1979)

suggesting higher marginal utility of incremental income in the loss domain. Given the

loss-aversion, individuals who are below their reference points may make more signi�-

cant attempts to rise the existing outcome than those above the reference point. People

consequently adopt di¤erent actions according to where a given outcome stands in rela-

tion to their reference values; their position in relation to a "reference value" determines

their motivation to work, undertake challenges, exert e¤ort. On a sample of physicians,

Rizzo and Zeckhauser (2003) for example show that individuals are inclined to work more

hours and take on additional tasks when their income is below their reference value (p.

915). Georgellis et al. (2008) study the adjustment towards reference wages and job

conditions for German workers. The authors show how the people�s inclination to change

their current job relates to their desire to reach the reference pay; these adjustments are

asymmetric and depend on the distance from the reference point and gender (Georgellis

et al., 2008). Adams (1963), Akerlof and Yellen (1990) and Krueger and Mas (2004)

argue that reference point in�uence people�s incentive to exert e¤ort. A worker that is

paid less than what she perceives as fair, is likely to reduce her e¤ort to re-establish her

pay-to-e¤ort ratio to a comparable level.

For an executive, the reference point is actually the benchmark, against which she

evaluates her compensation. The reliance on the executive pay surveys and peer-group

5

comparison to determine the benchmark for CEO compensation is today a common prac-

tice. In their review of the compensation reports of a hundred �rms in the S&P index,

Bizjak et al. (2007) for example report that 96% of sample �rms rely on peer group com-

parison in determining management compensation. Given the way they are constructed,

these benchmarks rarely signal the labor market conditions but indeed re�ect the CEOs�

beliefs about what they should be paid. The industry comparison in the executive pay

setting is in fact not neutral but often in�uenced by the CEO herself (Porac et al., 1999).

In fact, �rms rarely report what exactly constitutes their peer groups in the compensation

setting; for those who do report, the peer group for compensation often di¤ers from the

one used in the analysis of company performance. Consequently, as argued by Bebchuck

et al. (2001), a seeming objectivity in conformity with reference groups may limit the

optimal contracting due to subjectivity of a comparison base.

The question we are addressing in this paper relates to the issues discussed above. In

1997, the Association of Managers in Slovenia drafted a document, with which they aimed

to set the guidelines on what constitutes an appropriate remuneration for a Slovenian

executive. In the bargaining setting of Compte and Jehiel (2003), we can think of these

guidlines as the outcome of the �rst stage of the bargaining between the executives

and, the public. In fact, given that they were de�ned at the country level and publicly

discussed, the guidelines and the proposed pay levels already contained a consideration

of the potential outcry that the executives may cause if they were too demanding in

their claims. Still, the proposed pay levels were set well above the actual pay levels

of the Slovenian executive at that time; in fact, with these guidelines (also Criteria on

the executive pay or simply, Criteria), the Slovenian executives attempted to avoid pay-

losses in the economic turmoil after the secession from Yugoslavia and to preserve their

shares in the corporate rents. The aim of the discussions between the managers and

the representatives of other interest groups (i.e. the Association of the employers; the

Workers�Union, etc.) that surrounded the drafting and adoption of the guidelines was in

fact also to in�uence the public perception of what constitutes a proper remuneration and,

to re-anchor the �rm-level bargaining at a higher level. We would consequently expect to

observe adjustments in the executive compensation towards the values proposed in the

Criteria. In the continuation of the paper, we refer to these value as the "reference pay".

6

3 Reference pay

Slovenian executives were prior to transition considered as the top-end employees; their

�xed compensation was consequently a part of the general wage pool. The basic CEO

wages were determined on the country level, while the cross-sectional di¤erences in the

salaries mostly re�ected the di¤erences in education and job characteristics. By the

establishment of the Association of Managers at the beginning of transition (1989), the

Slovenian executives created a new interest group and legitimized their di¤erentiation

from other workers. One of the main moves of the Association was a proposal to rise the

executive pay in the largest �rms to a basic value of 1:5 in relation to the average employee

pay and, a modest performance bonus. A gradual rise of executive pay and pertaining

perks received sharp public protests, which re�ected the (still) egalitarian social values in

the country and the prevailing convictions that managers should not be treated di¤erently

from other workers.

The corresponding political interests stimulated discussions in the parliament and

ended up in the proposal to set an upper limit for executive pay and to provide comparable

pay levels across di¤erent �rms, especially those in substantial government ownership.

To prevent the adoption of a rigid regulation, the Association of Managers prepared the

Criteria on the executive pay. In this document, the top managers stipulated what they

considered the "appropriate" structure and level of their remuneration. The Criteria were

recognized also by the Chamber of Commerce, Chamber of Craft and Small business and

by the Association of Employers in 1994. In 1997, the document got published in the

O¢ cial Gazette and adopted the form of professional self-regulation. Despite the fact

that its adoption was discretionary, it did represent a strong "change" in the executives�

reference points. The guidelines introduced a substantial change in the de�nition of a

"fair" CEO pay and raised the multiplier of the average wage to 4 for small �rms, 6 for

medium-size �rms and to 8 for large �rms (before 5) 2. These levels of �xed pay could

be additionally increased by a maximum of 25% if a �rm outperformed the industry

or conversely, proportionately decreased. The executives could also be paid a bonus,

contingent on meeting speci�ed performance targets (maximum 30 percent). In this

regard, the Criteria recommended a limited set of performance measures, such as net

2The classi�cation of �rms to di¤erent size-groups followed the de�nitions �rms size of the SlovenianCompany Act (1993).

7

earnings, increase of exports and increase or retention of employment level, return on

equity (ROE) or on assets (ROA), market value and value added per employee. However,

no guidance was provided in relation to the weights that the �rms should attach to

a speci�c benchmark. Also, bonus was to be paid out of �rm pro�ts; due to double

taxation, this was however quite an unattractive type of compensation. In the late 90-

ties, cash bonuses in fact represented only between 13-15 percent of total executive pay

(Zupan, 1999; Merkaµc, 1997; Slapniµcar, 2002). The Criteria also de�ned a list of luxurious

non-quanti�able fringe bene�ts, a provision for severance payments and some guidance

for option compensation.3.

The Criteria on the executive pay were drafted in a very speci�c time of Slovenian

transition. The conclusion of the privatization process in the mid-90ties brought the �rst

real owners to Slovenian enterprises. This change implied a gradual re-distribution of

power from managers and employees to the new owners. The Criteria can be in fact

viewed as one of the ways through which the Slovenian managers tried to in�uence this

re-distribution to their own bene�t. As stated by one of the constitutive members of the

Association: �the Criteria were designed to guarantee an appropriate pay in the times of

�nancial distress....There is no pay limit for a good manager. The only upper limit for

the pay is its public acceptance. We need to actively in�uence this acceptance by proposed

pay levels. As pay ratios are now larger than before, we need to keep reconciling ours and

public views on the subject as long as the new pay ratios are not perceived as appropriate

and fair.� (Piskar, 2004, p. 19).

4 Empirical model speci�cation

This section provides an outline of empirical model that is used to tests the e¤ects of the

introduction of the Criteria on the CEO pay dynamics. We �rst describe the reference

pay as determined in the Criteria and then derive the empirical model. Thus, the rec-

ommended cash-compensation for CEO in �rm i in period t is limited to the following

3For instance, they suggested that the option exercise price should not be lower than the averagestock price from the preceeding year with no adjustments for market return. During the period of ouranalysis the average annual return of the Slovenian Stock Market Index 25.8 percent per year.

8

interval:

RefPayminit = 0:75WEcon;t +Wit

2� SMit (1)

RefPaymaxit = (1:25WEcon;t +Wit

2� SMit) � 1:3 (2)

where �WEcon;t denotes the average gross wage in the economy, �Wit is the average gross

wage in the �rm, and SMit is the �rm size multiplier. This multiplier is equal to 4

for small �rms, 6 for medium size �rms and 8 for large �rms. A rule that relates the

executive pay to �rm size and average wage in a �rm may be dynamically e¢ cient since

productivity and quality improvements allow a �rm to increase executive pay through the

increase of the average wage of the employees. Productivity improvements are essential

for �rm competitiveness, pro�tability and growth. Nevertheless, the relationship between

the executive pay and �rm size, a proxy for complexity of managerial work, was diluted by

the fact that Criteria introduced only a rough distinction between three size classes. This

rule, for example, o¤ers no incentives for the executives already working in large �rms.

An additional diluting e¤ect was in place due to the provision that the executive pay

should be partly related to the average wage in the economy. According to this rule, an

executive is not rewarded for performance of the �rm or the complexity of its operations,

but on factors over which she has no in�uence. Such a component provides a pay premium

in all the �rms that have a below- average productivity and penalizes those with above

average productivity levels. The correction for this problem was provided by a lump sum

increase or penalty of �xed pay rule (�25%) based on relative �rm performance. Hence,

the reference pay for the �xed part (RefPayfixit ) of the executive pay is determined by

the following formula:

RefPayfixit =�WEcon;t + �Wit

2� SMit; (3)

Total reference pay, on the other hand, considers also the impact of performance and

continuous e¤ect of �rm size and can be written as:

RefPaytotit = RefPayfixit (

1 + �it�11 + �medt�1

)��(litlmedt

)�l ; (4)

where � denotes the rate of return on assets and l is a number of employees. ��and �l

capture the elasticities of the actual pay on reference pay, size and performance measures.

9

The rate of return on assets is lagged for one period as bonuses are paid in April of

subsequent year and thus included in personal income tax statements for the next period.

Corresponding variables with superscript med denote industry-speci�c median values

(�rm i is in industry j). We use median values since these are less sensitive to extreme

values than the simple unweighted industry average. The simplest possible test of the

relevance of the reference pay�s impact on the managerial pay is to study the alignment

of the actual pay and the reference pay for the newly appointed executives in the �rst

year after appointment. By construction, this test does not allow an elimination of

time invariant person and �rm unobserved e¤ects. However, in order to control for the

observable �rm and individual characteristics, we introduce �xed e¤ects for organizational

form: stock corporation vs. closely held corporation (Dpublic), time (D time) and industry

(D industry), and individual characteristics (P), such as age and educational attainment.

We also allow for di¤erent responses of actual pay to reference pay between �rms with

di¤erent organizational forms and include an interaction term between reference pay and

the dummy for stock corporations �rms (Dpublic). In a log-linear form, the actual pay of

manager k in �rm i and period t is:

lnActPaykit = � lnRefPayfixit + �Public lnRefPayfixit DPublic;i (5)

+��(�it�1 � �medt�1 ) + �l(ln lit � ln lmedit ) +

+�PublicDPublic;i + �P lnPki + �Industry;jDij + �Time;sDist + "it;

where � denotes the elasticity of actual pay to reference pay. A correlation between the

average employee pay and the executive pay would cause the estimated elasticity � to

be positive. Note however that, upon the positive elasticity, we can not conclude that

the executive pay is determined in line with the reference pay. Only � equal to 1 can

be taken as a proof that executive pay is determined in accordance with the Criteria.

If this hypothesis holds for both organizational forms, the coe¢ cient that captures the

di¤erential e¤ect of reference pay for stock corporations (�Public) will be equal to 0. Both

of these hypotheses are tested below.

The pay contracts for existing managers are normally stipulated for a longer (3-4 year)

period. Consequently, the �rms are unlikely to adjust the compensation of the incumbent

managers to reference pay immediately. Hence, we specify that these adjustments take

10

place gradually. In this we follow Canarella and Nourayi (2008), who model gradual

adjustment of the executive compensation in 549 US �rms. Our speci�cation, however,

di¤ers from theirs in three important ways. First, we specify an adjustment dynamics

that relates the change in the actual pay to the change in the reference pay and the

distance between the reference pay and the actual pay. Second, we allow for asymmetric

adjustment of actual pay to positive and negative shifts in �rm size and positive and

negative distance between the actual and the reference pay. Third, we test whether the

executive pay setting di¤ers between the �rms with di¤erent organizational forms.

In particular, we specify the adjustment of executive pay for incumbent managers in

the following way4:

� lnActPaykit = �1� ln �Wit + �2� lnRefPaytot;ssit + �3(PayGapit�1):

Here � ln �Wit is the proportional change in the average wage of full time employees in the

�rm and � lnRefPaytot;ssit is the proportional change in the total reference pay due to the

change in the average wage in the economy, �WEcon;t; and the change in size multiplicator,

SMit. While each of these terms re�ects part of the change in the reference pay as de�ned

in Criteria, only the second term is consistent with the introduction of Criteria. Namely,

any rent-sharing rule that features simultaneous adjustment of compensation of employees

and executives, would yield a positive elasticity �1; whereas only the introduction of the

Criteria would yield also a positive �2. The last term on the right-hand side captures

the change in the actual pay due to closing of the gap between reference pay and actual

pay. PayGapit�1; denotes the lagged pay gap, de�ned as the di¤erence between the log

of total reference pay and the log of actual pay; �3 is the corresponding elasticity.

If all �rms set executive pay according to the Criteria in each period, all three elas-

ticities would be equal to 1. However, it is very likely that the CEO could more easily

adjust her pay according to the growth of average wage in a �rm, but much more di¢ cult

4Canarella and Nourayi (2008) study the following adjustment process of executive compensation tolong-run or optimal pay:

lnActPayit � lnActPayit = �(lnOptPayit � lnActPayit�1);

where OptPay denotes the long-run or optimal pay. Since they do not observe optimal pay, they cannot distinguish between contributions of change in the optimal pay and lagged deviation in optimal pay.Hence they can not study di¤erences in adjustment of actual pay to di¤erent components of the di¤erencebetween current optimal pay and lagged actual pay.

11

as a response to the pay gap, the shift in the �rm size or to the change in the average

wage in the economy. In line with previous �ndings, we expect that the larger the gap

between the reference and actual pay, the faster the adjustment to the reference pay. The

adjustments to close the pay gap are likely to be less frequent and as a consequence also

lower in e¤ect. For example, if only a quarter of �rms adjust to the pay gap in a given

period (every four years), �3 is reduced to one quarter. If reference points in�uence the

bargaining process, �2 and �3 would be di¤erent from zero. Given di¤erent underlying

forces, we assume that the strength of the three factors di¤er i.e. �1 6= �2 6= �3.

Theoretical arguments and previous empirical �ndings suggest that adjustments to

the reference points are asymmetric. For this reason we conjecture that the elasticities �2

and �3 di¤er depending depending on the shift of reference pay upwards or downwards

and on the actual pay being below or above the reference pay. In particular, we expect

greater elasticities for upward adjustments rather than for downward adjustments. Fol-

lowing Canarella and Nourayi (2008), we account for asymmetric adjustments with joint

estimation of the terms � lnRefPay and absolute value of the change in reference pay,

denoted by Abs(� lnRefPay). Similarly, for the asymmetric adjustments to pay gap we

account with lnPayGap and Abs(lnPayGap). Where lnPayGap > 0, the adjustment of

the pay gap in subsequent period is captured by the sum of coe¢ cients of lnPayGap and

Abs( lnPayGap); and where lnPayGap < 0; the adjustment is captured by the di¤erence

in coe¢ cients of lnPayGap and Abs(lnPayGap).

Thus far we have ignored the possibility that �rms with di¤erent level of managerial

power may exhibit di¤erent speeds of adjustment to the reference pay. According to

Bebchuk et al. (2002) the power of managers to increase their actual pay may depend on

the �rm governance characteristics, which in turn determine the managers�possibility to

in�uence the pay-setting negotiations. The most common variables that proxy the power

of the board vs. power of managers are the ownership structure, the composition of the

board and the tenure of the CEO. As none of these variables was available for our sample

in 1997, we proxy the shareholders�power with a dummy for the company�s legal form

(DCorp;i). The dummy adopts the value 1,when the CEO manages a stock corporation

and 0 when the CEO heads a limited liability �rm (closely-held �rm). At the time of

privatization, the legal form of stock corporation was mandatory for the �rms with at

least 50 shareholders; these were generally �rms with relatively more dispersed ownership

12

and control. We thus assume that in these �rms, the CEOs have a stronger bargaining

power vis-a-vis the boards than the CEOs in the limited liability �rms. This suggests that

the upward adjustments towards the reference pay should be relatively stronger, while the

downward adjustments should be relatively weaker in the stock corporations. However,

the legal form may as well exert the opposite e¤ects: the stock corporations are larger and

more visible, which makes their boards more sensitive to the public pressure concerning

the executive pay. Assuming that various constituencies (e.g. labor unions, general

public, ect.) believe that pay levels should be no more and no less than the "reference"

pay, the average speed of adjustment is likely to be higher for stock corporations on both

sides. For the downward pay adjustments, the latter hypothesis does not coincide with

the one based on the managerial power theory. We capture the e¤ects of the corporate

form on the positive and negative gap with the interaction term Abs(lnPayGap)�DCorp;i

as well as with the joint estimation of the interaction terms of � lnRefPay�DCorp;i and

Abs(� lnRefPay)�DCorp;i: The e¤ect of the dummy for the legal form is �xed over time

and hence not identi�ed in the �nal speci�cation.

In the speci�cation (4) we consider the total reference pay, which is unobservable. In

the empirical estimation we can only measure the �xed reference pay speci�ed in (3),

which di¤ers from the total reference pay due to the fact that, in constructing the �xed

reference pay, we are unable to account for the continuos change in the �rm size and,for

the performance e¤ects. Size is measured by the number of employees (l), whereas we

proxy the performance with the �rm�s deviation of return on assets from the industry

median (ROAit � ROAmedt ); these de�nitions correspond to what is stipulated in the

Criteria. Market performance is not considered due to the fact that a large share of our

sample consists of the �rms that are not listed on the Stock market. In the estimation

model, the performance is lagged for two years and �rm size for one year.5 Personal

characteristics (age, gender, educational attainment) of managers are excluded from our

analysis since they are �xed over time and hence not identi�able.

Based on all above, we estimate the following equation:

5Performance is lagged for two years because di¤erences of reference pay are estimated.

13

� lnActPaykit = �1� ln �Wit (6)

+�2�RefPayit +

+�3Abs�RefPayit +

+�4�RefPayit �DCorp;i +

+�5Abs�RefPayit �DCorp;i +

+�6(lnPayGapit�1) +

+�7Abs(lnPayGapit�1) +

+�8(lnPayGapit�1)�DCorp;i +

+�9Abs(lnPay Gapit�1)�DCorp;i +

+�10(ln lit � ln lmedit ) + �11(ln lit�1 � ln lmedit�1) +

+�12(�it�1 � �medit�1) + �13(�it�2 � �medit�2) +

+P

j2J �Industry;jDijt +P

s2T �Time;sDist +

+�ki + "kit;

where i = 1; :::N indexes the manager-�rm and t = 1; :::T indexes the time period.

D5 denotes the vector of industry dummies and year dummies. The nt represents the

time-speci�c term, the �ki is the manager-�rm speci�c �xed e¤ect and "kit is the random

disturbance.

5 Data

Data description

Testing our hypotheses before and after the adoption of Criteria in 1997 imposes signi�-

cant data requirements. For this purpose we merged �ve distinct data sets that contain

con�dential6 information on executive pay and publicly available accounting �rm-level

data. The identity of CEO for each �rm was established from the Statistical registry

of labor force (SRDAP), which contains employment records for all employees in each

6Public disclosure of executive compensation has been mandatory since 2002. However, the Law(Slovenian Companies Act, 2001) only requires the �rms to disclose the total compensation of the man-agement board but requires no individual disclosure for the CEO.

14

organization, including top managers with regular employment contracts.7 The employ-

ment record for each person contains information on occupation, and according to the

international standard classi�cation of occupations (ISCO) the top managerial position

has a unique code. The registry also contains personal characteristics of all employees,

such as age and educational attainment. Information on gross salaries of both CEOs

and employees was retrieved from the income tax records compiled by the Slovenian Tax

O¢ ce. Because of con�dentiality, we analyzed the data in a safe room at the Slovenian

Statistical O¢ ce. The Slovenian agency for public records (AJPES) collects the account-

ing information of the �rms. From this data source we use information on �rms�industry

a¢ liation (NACE). Slovenian Business registry (PRS) contains information on �rms�legal

forms of �rms.

We exclude micro and small �rms. In these �rms, the compensation practices follow

tax minimizing objectives and are less prone to rent extraction problems. We also exclude

�rms for which we could not �nd an employee with a CEO occupation code. Our data

base contains information on all relevant variables from 1995 to 2004; due to the use of

lagged variables in the estimations, we excluded the �rms, for which we do not dispose

with the data for at least three consecutive observations.

Summary statistics

This subsection provides basic descriptive statistics for the sample of �rms and CEOs used

in empirical analysis. The key �rm-level statistics are shown in Table 1. The average

�rm in our sample employs 325 employes, generates 29.1 million EUR of total sales and

operates with 35.1 million EUR of assets. The average value added per employee equals

23.2 thousand EUR. The average pro�tability is relatively low with Return on Assets

(ROA) equal to 2.2 percent and Return on Equity (ROE) of 5.8 percent.

For the purpose of our analysis, we classify the �rms in two main groups: the closely

held limited liability �rm (Ltd in the UK or GmbH in Germany) and stock corporations

(Plc in the UK or AG in Germany). Our sample is dominated by stock corporations

(Plc). Since �rms in the sample did not change the organizational form, this variable is

time invariant. The share of stock corporations is 63.5 percent in all �rms in the sample.

Most of the �rms (60 percent) belong to the mining, manufacturing and private utilities,

7CEO hired through an intermediary �rm can not be identi�ed since such �rms do not have a personwith code for top executive position.

15

followed by construction and trade.

Table 1: Descriptive statistics for �rms

Number of �rms 668Average number of observations per �rm 9.64

Size and Financial IndicatorsMean Std.Dev.

Employment 325 514Turnover 29.1 68.8Assets 35.1 85.1Value added per employee 23.2 19.7Debt to assets ratio 0.364 0.209Return on assets 0.022 0.072Return on equity 0.058 0.320

Ownership variablesAll years

Share of stock corporations 0.632

Sectoral structure of �rmsShare

Agriculture, Hunting, Fishing 0.017Mining, Manufacturing and Utilities 0.596Construction 0.112Trade 0.089Catering 0.028Transport and Communications 0.048Business Services 0.058Health Care 0.029Cultural and Recreational Services 0.024Total 1.000Source: AJPES, PRS and own calculations.

Notes: All nominal variables are given in euros, constant 2004 prices.

Annual turnover and sales are given in million euros.

Value added per employee is reported in thousand euros.

Table 2 reports the descriptive statistics for executives in the sample. The average

age was 46 years and the average share of women was 16.8 percent. More than 80 percent

of all executives held a University degree or master degree. 11.9 percent of managers

were replaced on average. The average gross salary for the CEOs that have been on

the position for the whole period of our analysis is 69.7 thousand EUR per year, with

16

standard deviation of 35.3 thousand EUR 8 The average executive pay was increasing

from 52.4 to 77.3 thousand euros (in constant 2004 prices), which suggests relatively low

salaries in comparison to other EU countries.

Table 2: Descriptive statistics for CEOs

Mean (Share) Std.Dev.Age 46.0 10.9Women 0.168 0.374University degree 0.816 0.388CEO turnover 0.119 0.323Gross annual salary 69.7 35.3Source: AJPES, PRS and own calculations.

Notes: Under University degree is reported the average

share of CEOs with at least 3-year undergraduate degree.

Tenure is reported from 1999 onwards.

Gross annual salary is reported in thousand euros (constant 2004 prices).

Table 3: Annual Salaries for CEOs

All CEOs Incumbent CEOsYear Mean Std.Dev. Mean Std.Dev.1995 52.4 20.4 53.6 20.81996 56.5 23.6 57.7 23.61997 61.5 26.7 62.3 27.01998 66.3 28.7 67.3 28.91999 69.8 31.6 72.0 31.22000 71.8 32.5 73.9 33.02001 73.5 39.1 75.0 40.12002 74.6 39.0 75.9 40.02003 75.4 40.7 77.9 41.82004 77.3 45.8 79.5 47.3Average 66.8 34.4 69.7 35.3Source: AJPES, PRS and own calculations.

Notes: Incumbent CEOs were employed in periods t and t-1.

Gross annual salary is reported in thousand euros (constant

2004 prices).

Next, we illustrate the adjustment of the actual pay to the "�xed" reference pay,

which was introduced in the 1997. Table 4 and Figure 5 document the dynamics of the

8The annual gross salary includes annual bonus.

17

ratio between the actual and the reference pay for the two organizational forms. Note

that we calculated the "�xed" reference pay for each �rm by using the annual gross wages

of all full time employees, the average gross wage in the economy and the size class, as

determined by the total assets, annual sales and total number of employees (including

part-time workers). Since we use annual data, these ratios are typically underestimated

for the managers that were replaced and thus have an incomplete employment spell in a

given year. For this reason, we use only data for the "incumbent" executives, namely for

the executives that were in executive position in period t� 1, t and t+ 1.

Table 4 shows that the average ratio between the actual and the reference pay in-

creased over time. However, in the limited liability �rms (closely held �rms), the actual-

to-reference pay ratio increased only by 10 percentage points, from 0.677 in 1996 to 0.778

in 2003, whereas in the stock corporations this ratio increased by 27 percentage points,

from 0.728 to 0.995. The observed pattern is further illuminated in the density plots of

the log of ratio between the actual and the reference pay (see Figure 5 below).9 Com-

parison of distributions over time con�rms a modest right-hand shift for closely held

�rms, whereas the distribution for the stock corporations shifted considerably, suggesting

greater responsiveness of actual to reference pay in this group �rms.

Table 4: Dynamics of average actual to reference pay ratio, 1996-2003

Year Closely held �rms CorporationsYear Mean Std. Dev. Mean Std. Dev.1996 0.677 0.192 0.728 0.2011997 0.697 0.226 0.769 0.2341998 0.719 0.234 0.827 0.2411999 0.720 0.282 0.868 0.2632000 0.720 0.280 0.865 0.2512001 0.727 0.375 0.876 0.2992002 0.802 0.325 0.996 0.3332003 0.778 0.351 0.995 0.315Source: AJPES, SORS, KDD, PRS and own calculations.

Notes: Statistics are calculated for incumbent CEOs.

9The densities for the log of actual-to-reference pay ratios are estimated with the method of stochastickernels. This method is convenient when the total number of observations is not large. This nonpara-metric method for plotting distributions generates smooth graphs. The method evaluates each point ofthe estimated density as a weighted sum of the data frequencies in the neighborhood of the point beingestimated. In our case the weighting is a normal (gaussian) density. The bandwidth around the pointof evaluation is 0.15. The larger is the bandwidth, the smoother is the estimated density. However, forour data, the qualitative features of the data are largely independent of selected bandwidth.

18

0.5

11.

5D

ensit

y

­1 .5 ­1 ­.5 0 .5 1

Ln Rat io  o f Act ual t o  Reference P ay

1 9 9 6  P u b l i c 2 0 0 3  P u b l i c

1 9 9 6  P riv a te 2 0 0 3  P riv a te

So u rce:  SO RS an d  o w n  calcu lat io n s .

6 Empirical results

In this section we present the estimates of the empirical models for the newly appointed

and incumbent executives. We start with estimation of the pay equation for the newly

appointed executives (5). The method of estimation is OLS on a pooled panel of manager-

�rm observations, since we can treat the lagged endogenous variables (e.g. labor produc-

tivity, employment, return on assets) as predetermined.

We present the estimates of the pay equation in Table 5. The key coe¢ cient of interest

is the elasticity of the actual pay to �xed reference pay. Failure to reject the null of

� = 1, con�rms determination of executive pay according to the Criteria. The estimated

elasticities for the limited liability �rms and stock corporations are 0.725 (s:e: = 0:14) and

0.914 (0:725+0:199, s:e: = 0:15) respectively. The null of � = 1 cannot be rejected only for

the case of stock corporations; thus, we can conclude that in these �rms, the compensation

of the newly appointed directors follows the reference wage, which was stipulated in the

Criteria on the executive pay. Reference pay to a large extent incorporates the e¤ect

of size; however, it is not captured entirely since the construction of the reference pay

accounts only for the three size brackets (i.e. small, medium and big �rms). Consequently,

we observe a positive and signi�cant impact of the level of �rm employment on the total

19

CEO pay. The corresponding coe¢ cient suggests that an increase in the �rm size by 10

percent (over the industry median) increases the total pay by 1.08 percent. In addition,

we �nd that more productive �rms tend to give larger rewards to CEOs, while this

is not the case for relatively better performing �rms (see the insigni�cant sign of the

ROA coe¢ cient). The impact of the �rm productivity on the total pay is of the same

magnitude as for the size; it must be however noted that the coe¢ cient for productivity

(see the variable Log of value added per employee in Table 5) does not capture the total

productivity e¤ect; a higher productivity allows the �rms to pay higher wages to their

employees, which re�ects in the higher reference wage. Finally, the level of CEO pay does

not depend on the legal form of the �rm, neither on the personal characteristics (age

and education) of the CEO. To sum up, reference points are shown to be an important

determinant of the pay for the newly appointed CEOs in the stock corporations. In the

closely held �rms, we are not able to determine whether the increases in the CEO pay

re�ect the reference wage or are simply the result of a general wage increase in a given

�rm. Rather than the in�uence of a new reference value of the compensation contracts,

the latter e¤ect re�ects the rent-sharing between the managers and the employees (i.e.

the distribution of the newly created value added through higher wages of both managers

and workers).

In the next step, we estimate the dynamic pay equation for the incumbent managers

(6). In these estimation, we only include the CEOs that were in the �rm at least in the

periods t � 2, t � 1, t and, t + 1. The latter restriction is necessary to obtain correct

measures for the annual growth rates of the CEO compensation. To ensure that the pay

in year t refers to the entire year, the CEO�s presence in period t + 1 is required. In

other words, we only consider �rms where the same manager remained on the position

in the consecutive year to which the change of pay refers. The results of the Blundell-

Bond (1998) estimations of the pay di¤erence equation are presented in Table 6 below.

The Sargan test of the over-identifying restrictions con�rms the global validity of the

instruments employed in the estimation. The Arellano-Bond test for the second-order

serial correlation in the �rst-di¤erenced residuals also con�rms the absence of the second-

order autocorrelation. Standard errors are robust for heteroskedasticity; this correction

is needed as the pay dispersion varies with the �rm size.

The results of this �nal estimation generate some interesting conclusions. First, the

20

Table 5: Estimates of Actual Pay Equation for New CEOs

Coe¢ cientVariable (s.e.)

Reference payLog of Reference Pay 0.715***

(0.14)Log of Reference Pay � DCorporation 0.199

(0.16)CEO characteristicsLog of Age 0.106

(0.072)DHigherEducation 0.00582

(0.056)

Firm characteristicsLog of employmentt�1 0.108**

(0.036)Log of value added per employeet�1 0.108***

(0.022)ROAt�1 -0.159

(0.20)DCorporation -2.096

(1.77)Constant 2.148

(1.64)

Observations 353Time and industry dummies YESR2 Adjusted 0.478F(22,330) test 15.66p-value 0.000Source: AJPES, SORS, KDD, PRS and own calculations.

Robust standard errors in parentheses. *** p<0.001, ** p<0.01, * p<0.05

Log of employment, log of labor productivity and ROA are included as

deviations from industry median values.

strongest trigger of the CEO pay increases is the rise of the average employee wage

� ln �Wit at the �rm level. The coe¢ cient amounts to 0.514 and is highly signi�cant.

This result indicates that the employees partly condition the increases in CEO salaries

by the level of their own salaries. That is, through the salary increases, the managers

and the employees jointly gain a share of the value added, which is generated in the

�rm (rent-sharing). The e¤ect of the reference pay is estimated through the variable

21

�RefPayit;by including the term Abs�RefPayit; we control for a possible asymmetry

in the adjustments in the case of a decrease in the reference pay. For limited liability

corporations, the impact of RefPay is 0.128 and not signi�cant. We calculate the value

of the coe¢ cient as the sum of the coe¢ cient for �RefPayit and the Abs�RefPayit: For

decreases of RefPay, we refer to the di¤erence of the two coe¢ cients; the �nal coe¢ cient

is 0.215 (0.104210) and signi�cant. The results suggest that, in the closely held �rms,

the CEO compensation is in�uenced by the reference values only when the latter de-

crease; that is, we only observe adjustments downwards and not upwards. For the stock

corporations, the e¤ect of a change in the reference pay is estimated by summing up

�RefPayit; Abs�Re fPayit;�RefPayit �DPublic;i and Abs�RefPayit �DPublic;i: The

�nal coe¢ cient is signi�cant and amounts at 0.2671 (0.072). Thus, the executive pay in

the stock corporations is much more sensitive to the increases of the reference pay. The

adjustment downward (0.161) is however not signi�cant, implying that the CEO salaries

in these �rms may be rigid downwards. Finally, how does the distance from the reference

pay in�uence the pay adjustments? In this regard, we obtain the following results: the

impact of prior period pay gap for managers of closely held �rms,whose pay is below

the reference point, is estimated at 0.120 (0.046) and signi�cant. In the case of stock

corporations, the impact of the gap is stronger, 0.249 (0.045) and also signi�cant.

For managers that already earn more than the reference pay in the closely held �rms,

the gap has no e¤ect on the pay adjustments. The downward rigidity of the CEO com-

pensation is con�rmed also for the case of stock corporations; the gap coe¢ cient amounts

to 0.188 (0.098) but is not signi�cant. Large standard errors for the latter coe¢ cients

are due to the low number of the observations since only a few managers actually earn

more than the reference value (during the period of our analysis). These results con�rm

what can be generally observed from our descriptive statistics: the upward response to

the pay gap is stronger for stock corporations than for the limited liability �rms, while

downward adjustments are rigid for both groups of companies.

We also �nd that an increase in size in year t has a signi�cant and positive e¤ect on

the pay growth; the e¤ect of (t-1) is not signi�cant. An increase of ROA provides no

contribution to the explanation of the pay rises. It must be noted that the performance

e¤ect is partly incorporated into the rise of the average employee wage at the �rm level.

10The brackets report the corresponding standard errors.

22

To sum up, our empirical analysis of the CEO pay leads to three main conclusions.

First, increases in the managerial salaries are to a large extent determined by the (cor-

responding) increases in the employee wage, which con�rms that managers share part

of their rents with the employees. Secondly, reference pay matters; we observe that the

actual CEO pay adjusts in line with the changes in the reference pay and that the CEO

pay grows stronger in the �rms, where the CEO�s compensation is more distant from

the reference value. These adjustments are stronger in the stock corporations, where

the ownership is more dispersed and consequently, the managers have a stronger in�u-

ence over their compensation; the CEO salaries in these �rms are also rigid downwards.

These �ndings support the conclusion that reference values matter more in the �rms with

stronger managers (and weaker owners).

7 Conclusions

A number of theoretical and empirical evidence shows that people�s actions can be partly

explained by their desire to reach some reference income, which they consider as the

fair compensation for their e¤ort. Firm executives are subject to the same behavioral

patterns; it has been largely observed that reference values play an important role in

in�uencing the CEOs�perception of what they should get as the compensation for their

work. Peer-group comparison and industry surveys are just few examples of the fact

that the managers evaluate their pay in relation to certain benchmarks or, reference

points. This paper builds on the observation from the Western practice but relies on

a somehow di¤erent setting. We show the example of a new European country, where

in order to prevent their share in the corporate rents, the managers joined their forces

and self-determined the "reference" level for their compensation. We show how these

(new) reference values signi�cantly in�uenced the stipulation of the contracts for new

managers and also, how they induced signi�cant adjustments in the pay levels of the

existing (incumbent) managers. The adjustments were stronger in the stock corporations,

where (due to dispersed ownership) the managers had a stronger bargaining power against

the owners. These results and the facts surrounding the introduction of the new reference

value for the executive pay in Slovenia provide a country-example in support to the

managerial power theory of the executive pay.

23

References

[1] Adams, J. S. (1963), Toward an understanding of inequity. Journal of Abnormal and

Social Psychology, 67, 5: 422 �436.

[2] Anderson, T. W., Hsiao, C. (1981), Estimation of Dynamic Models with Error Com-

ponents, The Journal of American Statistical Association, 76: 598 �606.

[3] Akerlof, G. A., Yellen, J. L. (1990), The Fair Wage-E¤ort Hypothesis and Unem-

ployment, The Quarterly Journal of Economics, 55, 2: 255 �283.

[4] Arellano, M. , Bond, S. (1991), Some Test Speci�cation for Panel Data: Monte Carlo

Evidence and an Application to Employment Equations. The Review of Economic

Studies, 58:277-297.

[5] Bebchuck, A. L., Fried, J. M., Walker, D. I. (2001), Executive compensation in

America: Optimal contracting or extraction of rents? NBER Working Paper Series

8661, http://www.nber.org/papers/w8661.

[6] Bebchuk, A. L., Jesse, M. F., Walker, D. I. (2002), Managerial Power and Rent

Extraction in the Design of Executive Compensation, The University of Chicago

Law Review, 69, 3: 751 �846.

[7] Bebchuck, A. L., Fried, J. M. (2003), Executive Compensation as an Agency Prob-

lem. Journal of Economic Perspectives, 17, 3, 71 �92.

[8] Blundell, R., Bond, S. (1998), Initial conditions and moment restrictions in dynamic

panel data models, Journal of Econometrics, 87: 115 �143.

[9] Bizjak, J. M., Lemmon, M. L., Naveen, L. (2007), Does the Use of Peer Groups Con-

tribute to Higher Pay and Less E¢ cient Compensation, availabl at www.ssrn.com.

[10] Canarella, G., Nourayi, M., M. (2008), Executive Compensation and Firm Per-

formance: Adjustment Dynamics, Non-linearity and Asymmetry, Managerial and

Decision Economics, 29: 293 �315.

[11] Eriksson, T. (2005) Managerial Pay and Executive Turnover in the Czech and Slovak

Republics, Economics of Transition, 13, 4: 659 �677.

24

[12] Ezzamel, M., Watson, R. (1998), Market Comaprison Earnings and the Bidding-Up

of Executive Cash Compensation: Evidence from the United Kingdom. Academy of

Management Journal, 41, 2: 221 �231.

[13] Ezzamel, M., Watson, R. (2002), Pay Comparability Across and Within UK Boards:

An Empirical Analysis of the Cash Pay Awards to CEOs and Other Board Members.

Journal of Management Studies, 39, 2: 207 �232.

[14] Easterlin, R. A. (1995), Will Raising the Incomes of All Increase the Happiness of

All? Journal of Economic Behavior & Organization, 27: 35 �48.

[15] Falk, A., Fehr, E., Zehnder, C. (2006), Fairness Perceptions and Reservation Wages

- The Behavioural E¤ects of Minimum Wage Laws. The Quarterly Journal of Eco-

nomics, November: 1347 �1381.

[16] Farell, K. A., Friesn G., C., Hersch, P. L. (2008), How do Firm Adjust Director

Compensation?, Journal of Corporate Finance, 14: 153 �162.

[17] Georgellis, Y., Gregoriou, A., Tsitsianis, N. (2008), Adaptation Towards Reference

Values: A non-linear perspective, Journal of Economic Behavior & Organization,

67: 768 �781.

[18] Hart, O., Moore, J. (2006), Contracts as Reference Points, Working paper, available

at www.ssrn.com

[19] Holmstrom, B., Kaplan, S. (2003), The State of US Corporate Governance: What�s

Right and What�s Wrong? Journal of Applied Corporate Finance, 15: 8 �20.

[20] Jones, D., and Kato, T. (1996) The Determinants of Chief Executive Compensation

in Transitional Economies: Evidence From Bulgaria, Labor Economics, 3: 319 �336.

[21] Compte, O., Jehiel, P. (2003), Bargaining with Reference Dependent Preferences,

available at www.ssrn.com.

[22] Jensen, M. C., Murphy, K. J. (1990), Performance Pay and Top Management Incen-

tives, Journal of Political Economy 98: 225 �264.

25

[23] Jensen, M. C., Murphy, K. J. (2004), Remuneration: Where we�ve been, how we got

to here, what are the problems, and how to �x them, European Corporate Governance

Institute, Finance Working Paper N.44.

[24] Jones, D., and Kato, T. (1996), The Determinants of Chief Executive Compensation

in Transitional Economies: Evidence From Bulgaria, Labor Economics, 3: 319 �336.

[25] Kahneman, D. (2002), Maps of Bounded Rationality: A Perspective on Intuitive

Judgment and Choice, Nobel Prize Lecture, December 8, 2002..

[26] Kahneman, D., Tversky, A. (1979), Prospect theory: An analysis of decisions under

risk., Econometrica, 47: 313 �327.

[27] Koszegi, B., Rabin, M. (2006), A Model of Reference-Dependent Preferences, The

Quarterly Journal of Economics, 71, 4: 1133 �1165.

[28] Li, D. (2004), Bargaining with History Dependent Preferences, Boston University,

Mimeo.

[29] Merkaµc, M. (1997, Primerjava podatkov v strukturi plaµce, o razmerjih v plaµci, o

stro�kih dela in o individualnih pogodbah v slovenskih podjetjih (Eng. Analysis

of pay structure, pay ratios, costs of labour and individual managerial contracts in

Slovenian �rms). In (ed. Kejµzar) Procesi inovativnega razvoja organiziranosti in mo-

tiviranja kadrov. Proceedings from XII. conference. Bled: Dru�tvo za vrednotenje

dela, organizacijski in kadrovski razvoj, May, 129 - 138 (in Slovenian).

[30] Murphy, K. J. (1999), Executive Compensation, in Handbook of Labor Economics

(Orley Ashenfelter and David Card eds.). Amsterdam: North Holland.

[31] Perry T., Zenner, M. (2001): Pay for performance? Government regulation and the

structure of compensation contracts. Journal of Financial Economics, 62: 453 �488.

[32] Perel M. (2003), An Ethical Perspective on CEO Compensation. Journal of Business

Ethics, 48, ·4: 381�391.

[33] Porac, J. F., Wade, J. B., Polloc, T. G. (1999), Industry Categories and the Politics

of the Comparable Firm in CEO Compensation. Administrative Science Quarterly,

44: 112 �144.

26

[34] Rizzo, J., A., and R., J., Zeckhauser (2003), Reference Incomes, Loss Aversion, and

Physician Behavior, The Review of Economics and Statistics, November 2003, 85, 4:

909 �922.

[35] Rose, N. L. and Wolfram, C. (2000), "Has the "million-dollar cap" a¤ected CEO

pay?", American Economic Review, 90: 197 �202.

[36] Bertrand M., Mullainathan S. (2001), Are CEOs rewarded for luck? The ones with-

out principals are, Quarterly Journal of Economics, 116: 901 �932.

[37] Bruce, A., Buck T., Main, B. G. M. (2005), Top executive remuneration: a view

from Europe. Journal of Management Studies, 42: 1493 �1506.

[38] Slapniµcar, S. (2002) Managing Earnings: Factors that May In�uence the Choice of

Depreciation Method in Large Slovenian Firms. Economic and business review, 4, 2:

103 �127.

[39] Piskar, S. (2004), Fiftheen years since the establishement of the Association of Man-

agers, which we can be proud of. In (eds. Kra�ovec, T., Seme, B.), The Slovenian

Association of Managers from 1989 - 2004, 15 �20.

[40] Wade, J. B., O�Reilly, C. A. III., Pollock, T. G. (2006), Overpaid CEOs and Under-

paid Managers: Fairness and Executive Compensation. Organization Science, 17, 5:

527 �544.

[41] Wade, J. B., O�Reilly, C. A. III., Pollock, T. G (1990), Golden Parachutes: CEOs

and the Exercise of Social In�uence. Administrative Science Quarterly, 35:587-603.

[42] Zajac E. J., Westphal J. D. (1995), Accounting for the Explanations of CEO Com-

pensation: Substance and Symbolism, Administrative Science Quarterly, 40, 2: 283

�308.

[43] Zupan, N. (2000) Nagrajevanje vodilnih usluµzbencev (Rewarding of top manage-

ment). In (ed. Turk) Zbornik 32. simpozija o sodobnih metodah v raµcunovodstvu,

�nancah in reviziji (Proceedings from the Conference Contemporary Methods in

Accounting, Finance and Audit), Portoroµz, 22 - 24 March: 119 �133. (in Slovenian)

27

Table 6: Estimates of Pay Equation for Incumbent CEOs

Coe¢ cientVariable (s.e.)Growth of Average �rm wage 0.514***

(0.097)Reference pay and Pay gapGrowth of Reference Pay 0.174*

(0.078)Abs of Growth of Reference Pay -0.0455

(0.089)Growth of Reference Pay� DCorp;t�1 0.0403

(0.085)Abs of Growth of Reference Pay� DCorp;t�1 0.0988

(0.094)Log Pay Gapt�1 0.138*

(0.062)Abs of Log of Pay Gapt�1 -0.0173

(0.073)Log Pay Gapt�1� DCorporation 0.0808

(0.067)Abs of Log Pay Gapt�1� DCorporation 0.0477

(0.067)Firm characteristicsLog of employmentt�1 0.106***

(0.032)Log of employmentt�2 -0.041

(0.028)ROAt�1 0.161

(0.105)ROAt�2 -0.002

(-0.003)Constant -0.304

(0.17)Speci�cationObservations 2623Wald chi2(27) 418.5p-value 0.000AR(1) -9.522p-value 0.000AR(2) 0.104p-value 0.918AR(3) -0.881p-value 0.378Sargan chi2(250) 254.49p-value 0.409Source: AJPES, SORS, KDD, PRS and own calculations.

First step robust standard errors in parentheses. *** p<0.001, ** p<0.01, * p<0.05

Log of employment, log of labor productivity and ROA are included as deviations

from industry median values.

All right-hand side variables are included as lagged instruments. Instruments

for di¤erenced equation are lagged values t-1 and t-2. Instruments for the

level equation are di¤erences from periods t-1 and t-2.

28